Extensions to the p4p Server API, to aid in creation and managment of pvAccess servers.
p4pillon may be installed via
$ pip install p4pillonRequires Python 3.10 or later. These extensions make extensive use of typing and other recent Python features.
Documentation intended for developers using the library is available here.
A brief overview of the components of the library.
Caution
This is not an alternative to the Process Database implemented in a traditional EPICS IOC. Although the Normative Type logic is implemented, it does not implement locking. This means that in the case of multiple simultaneous updates it is possible for a PV to become inconsistent. At this time we suggest that the NT Logic code be used for rapid prototyping and systems where consistency/reliability are not critical.
The SharedNT class, derived from p4p's SharedPV class, automatically implements the logic of Normative Types (at this time NTScalars and NTScalarArrays) using handlers. It is possible, with the CompositeHandler class to override this behaviour, including diabling or replacing Normative Type logic.
p4p only allows a single Handler class to be associated with a SharedPV. To make it easier to combine or share handlers from multiple sources a CompositeHandler class is provided. The supplied CompositeHandler is derived from the p4p Handler. It also derives from an OrderedDict to store componenent Handlers (also standard p4p Handlers). The CompositeHandler calls component handlers in the specified order, and allows the component Handlers to be accessed through the OrderedDict interface.
Note
CompositeHandler is designed to work with the Handler class, and is not designed to work with the Handler decorators.
Rules are component Handlers used with the CompositeHandler used to implement the NT Logic discussed above. The Rule class is derived from the Handler class but implements a commonly encountered flow from put() (for identification and authorisation), to put() (for operations that require comparison between the current state of the PV and its possible future state), and open() (for operations that only need consider the possible future state of the PV).
PVRecipe is a factory method used to simplify creation of SharedNT objects, i.e. PVs that implement Normative Type logic. It may be used in conjunction with the p4pillon Server class to simplify management of event loops and PV lifecycle.
The config_reader parses a YAML file in order to construct SharedNT PVs which are managed by the p4pillon Server. This is the simplest way to create and configure
Install the extra dependencies required for testing using pip install .[test] or similar.
To run tests invoke pytest:
$ python -m pytest testsor to run all tests and output a coverage report:
$ uv run --extra=test python -m coverage run --source=. -m pytest -x tests
$ uv run --extra=test python -m coverage report
This repository's CI/CD pipeling (using GitHub Actions) checks that source code meets PEP 8, and other more stringent, coding standards. This uses the ruff linter and code formatter. It is included in the `.[test] dependencies (see above) and may be manually invoked:
$ ruff check --fix
$ ruff formatThe release process requires use of the .[dist] dependencies, which may be installed with pip install .[dist]. A build may then be triggered with python -m build. Alternatively, use:
$ uv run --extra=dist python -m buildPublication to either PyPi or TestPyPI is performed automatically via CI/CD (i.e. GitHub Actions) and is driven by tags. Any commit intended for package publication must be tagged with a unique tag, and the semantic version must be greater than any existing tag.